A method and device for monitoring damage of vibrating screen based on deep learning

Abstract

The present invention relates to the field of concrete mixing technology, and in particular to a method and device for monitoring the damage of a vibrating screen based on deep learning. A method for monitoring the damage of a vibrating screen based on deep learning, wherein an image acquisition platform is built above a single-stage unloading chute after the machine-made sand is filtered through the screen, and images of stacked machine-made sand in the falling process are collected; the images collected before and after the screen is damaged are made into a binary classification data set, which are input into a lightweight convolutional neural network for training to obtain a classification model, and the classification model is deployed on an edge computer to build an online monitoring system; the system analyzes the collected images in real time, and when the number of images after the screen is damaged reaches a certain number, the system issues a screen damage alarm. The present invention can realize the tracking and monitoring of the status of the vibrating screen, intuitively grasp the degree of damage to the screen, and replace the damaged screen in time to ensure the production quality of the machine-made sand.

Description

Technical Field

[0001] The present invention relates to the field of concrete mixing technology, and in particular to a method and device for monitoring damage of a vibrating screen based on deep learning. Background Art

[0002] As a green building material, machine-made sand has absolute advantages in environmental protection and cost, and is widely used in concrete production. As a key component in sand making equipment, the main function of the vibrating screen is to effectively classify the machine-made sand after crushing to ensure that the final output particle size distribution can meet the engineering quality requirements. However, the vibrating screen mesh of the current sand making equipment often faces the problem of damage during use. This is because the mesh is subjected to high load for a long time, and a large amount of material impacts and scratches the mesh, which not only affects production efficiency, but also leads to a decline in the quality of machine-made sand products.

[0003] Due to the different hardness, humidity, particle size and other factors of the material, the wear speed and damage mode of the screen are also different. At present, it is usually set up for patrol personnel to sample the material and then observe and judge it with the naked eye. The manual inspection method has problems such as low efficiency, strong subjectivity and high cost. At the same time, the traditional manual inspection method cannot monitor the status of the screen in real time, and it is often discovered only after the screen is damaged to a certain extent. This not only increases downtime, but may also affect product quality due to the mixing of materials caused by the damaged screen. Summary of the invention

[0004] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or be understood by implementing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the description and other drawings of the description.

[0005] The purpose of the present invention is to overcome the above-mentioned shortcomings and provide a method and device for monitoring the damage of a vibrating screen based on deep learning, which can quickly identify the collected original image of machine-made sand and judge the state of the screen, thereby forming a real-time vibrating screen state monitoring system, which is conducive to tracking the state of the vibrating screen and timely alarm; at the same time, the intelligent monitoring system is deployed on an edge computing intelligent terminal, the device has a high degree of integration, simple installation, high portability, long non-contact monitoring life, and convenient maintenance; it replaces the traditional manual inspection method, which is convenient for equipment management personnel to maintain the vibrating screen in time, ensure product quality, improve production efficiency, reduce operation and maintenance costs, and generate good economic benefits.

[0006] The present invention provides a vibration screen damage monitoring method based on deep learning, comprising:

[0007] S1. Using a vibrating screen to screen the machine-made sand, the machine-made sand after screening falls into the material discharge chute below, and using an image acquisition platform to collect images of the machine-made sand in the material discharge chute. The machine-made sand images specifically include: images of single-grade materials before the screen is damaged, and images of mixed materials of machine-made sand with a larger particle size after the screen is damaged;

[0008] S2, screening the obtained machine-made sand images and making a binary classification data set;

[0009] S3, input the binary classification data set into the lightweight convolutional neural network for training to obtain a classification network model;

[0010] S4. Deploy the trained classification network model on the edge computing device, build an online monitoring system in combination with the image acquisition platform, and monitor the vibrating screen in real time;

[0011] S5. Process and analyze the collected machine-made sand image in real time. When the system determines that the screen is damaged and the particle size in the machine-made sand image reaches a specified threshold, the system issues a screen damage alarm.

[0012] In some embodiments, in step S1, the machine-made sand sieved by the complete vibrating screen is machine-made sand of a single particle size grade, the material discharge chute is inclined, and the image acquisition platform is arranged perpendicular to the material discharge chute.

[0013] In some embodiments, in the process of converting the machine-made sand image into a binary classification data set in step S2, a data enhancement step is also required. The specific methods of data enhancement include: horizontal flipping, vertical flipping, brightness change, adding noise and blurring.

[0014] In some embodiments, in step S3, the lightweight convolutional neural network adopts the MobileNetV3 network structure, a large version of which includes 15 bottleneck layers, one standard convolution layer and three point-by-point convolution layers, and the bottleneck layer includes multiple depth-separable convolution layers.

[0015] In some embodiments, the classification network model is a deep learning framework based on PyTorch, and the TensorRT inference engine is used to accelerate the inference of the classification network model.

[0016] In some embodiments, in step S4, the core module of the edge computing device is an NVIDIA Jetson XavierNX embedded artificial intelligence computing processor.

[0017] In some embodiments, in step S5, the specific manifestations of the system issuing a screen damage alarm include: displaying the current machine-made sand image, prompt information displayed synchronously with the machine-made sand image, and sound and light alarm information running synchronously with the sound and light alarm.

[0018] A vibrating screen damage monitoring device based on deep learning, comprising:

[0019] A screening device, in which a plurality of vibrating screens of different densities are arranged;

[0020] A material discharge chute connected to the discharge port of the screening device, and the material discharge chute is inclined;

[0021] An image acquisition platform is disposed above the material discharge chute, and the image acquisition platform is perpendicular to the material discharge chute;

[0022] An edge computing device electrically connected to the image acquisition platform;

[0023] An audible and visual alarm is arranged on the screening device and is electrically connected to the edge computing device.

[0024] In some embodiments, a material distributor is provided in the discharge chute at a position corresponding to the image acquisition platform, and the material distributor raises the height of the machine-made sand falling into the discharge chute.

[0025] In some embodiments, the edge computing device includes a memory and a processor, the memory stores a vibrating screen damage monitoring computer program, and the computer program is loaded and executed on the processor.

[0026] By adopting the above technical solution, the beneficial effects of the present invention are:

[0027] The vibrating screen damage monitoring method based on deep learning proposed in the present invention can quickly identify the collected original image of machine-made sand and judge the status of the screen, thereby forming a real-time vibrating screen status monitoring system, which is conducive to tracking the status of the vibrating screen and timely alarm; at the same time, the intelligent monitoring system is deployed on the edge computing intelligent terminal to run, the device has a high degree of integration, simple installation, high portability, long non-contact monitoring life, and convenient maintenance; it replaces the traditional manual inspection method, which is convenient for equipment management personnel to maintain the vibrating screen in time, ensure product quality, improve production efficiency, reduce operation and maintenance costs, and generate good economic benefits.

[0028] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure.

[0029] Undoubtedly, these and other objects of the present invention will become more apparent after the following detailed description of the preferred embodiment described with reference to various figures and drawings.

[0030] In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, one or several preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention and do not constitute a limitation of the present invention.

[0032] In the drawings, the same reference numerals are used for the same components and the drawings are schematic and not necessarily drawn to scale.

[0033] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only one or several embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on such drawings without paying creative work.

[0034] Figure 1 It is a schematic diagram of the overall flow of the monitoring method in some embodiments of the present invention;

[0035] Figure 2 A schematic diagram of an original image captured by an image capture platform in some embodiments of the present invention;

[0036] Figure 3 A schematic diagram of the structure of the image classification network MobileNetV3 in some embodiments of the present invention;

[0037] Figure 4 Schematic diagram of the overall structure of the monitoring device in some embodiments of the present invention.

[0038] Description of main reference numerals:

[0039] 1. Screening device; 2. Unloading chute; 3. Image acquisition platform; 4. Edge computing equipment; 5. Sound and light alarm; 6. Material divider. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, but are not used to limit the present invention.

[0041] In addition, in the description of the present invention, it should be understood that the terms "center", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "axial", "radial", "circumferential" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present invention.

[0042] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral body; it can be directly connected or indirectly connected through an intermediate medium, and it can be the internal connection of two elements or the interaction relationship between two elements. However, if it is indicated as a direct connection, it means that the two connected bodies are not connected through a transition structure, but are connected to form a whole through a connecting structure. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to the specific circumstances.

[0043] In the present invention, unless otherwise clearly specified and limited, the first feature "on" or "under" the second feature may be that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in an appropriate manner in any one or more embodiments or examples.

[0044] Reference Figure 1-Figure 4 , Figure 1 It is a schematic diagram of the overall flow of the monitoring method in some embodiments of the present invention; Figure 2 A schematic diagram of an original image captured by an image capture platform in some embodiments of the present invention; Figure 3 A schematic diagram of the structure of the image classification network MobileNetV3 in some embodiments of the present invention; Figure 4 Schematic diagram of the overall structure of the monitoring device in some embodiments of the present invention.

[0045] According to some embodiments of the present invention, the present invention provides a method for monitoring damage of a vibrating screen based on deep learning, comprising:

[0046] S1, such as Figure 4 As shown, a vibrating screen is used to screen the machine-made sand, and the machine-made sand after screening falls into the material discharge chute 2 below. The image acquisition platform 3 is used to collect images of the machine-made sand in the material discharge chute 2, as shown in FIG. Figure 2 As shown, the machine-made sand image specifically includes: a single-grade material image before the screen is damaged and a mixed material image of machine-made sand with a larger first-grade particle size after the screen is damaged;

[0047] The machine-made sand sieved by the complete vibrating screen is machine-made sand of a single particle size grade. The material discharge chute 2 is inclined, and the image acquisition platform 3 is arranged vertically to the material discharge chute 2 .

[0048] S2, such as Figure 1 As shown, the acquired machine-made sand images are screened and made into a binary classification data set; a data enhancement step is also required, and the specific methods of data enhancement include: horizontal flipping, vertical flipping, brightness change, adding noise and blurring.

[0049] S3, input the binary classification data set into the lightweight convolutional neural network for training to obtain a classification network model;

[0050] like Figure 3 As shown in the figure, the lightweight convolutional neural network adopts the MobileNetV3 network structure, in which the large version includes 15 bottleneck layers, one standard convolution layer and three point-by-point convolution layers. The bottleneck layer contains multiple depth-separable convolution layers.

[0051] MobileNetV3 combines the platform-aware NAS and NetAdapt algorithms to automatically search for the optimal network structure. The bottleneck layer is the basic building block of MobileNetV3. Each bottleneck layer mainly implements channel-separable convolution, optional SE channel attention mechanism, and residual connection. Specifically, first, channel compression is performed through 1x1 convolution, that is, dimensionality reduction operation, and then deep separable convolution is performed to extract features, such as 3x3 or 5x5 convolution, the number of groups is equal to the number of channels, and finally 1x1 convolution is used to aggregate channels, that is, dimensionality increase operation; Among them, MobileNetV3 uses h-swish activation function;

[0052] The training strategy specifically includes: loading the pre-trained model obtained by training the large-scale dataset ImageNet to initialize the network weights of MobileNetV3 before starting training; dividing the produced dataset into training set, validation set and test set according to 7:2:1 for training; the training process adopts the gradient descent optimization algorithm SGD, the learning rate automatic adjustment strategy, and the learning rate is adjusted every 30 epochs, and the learning rate decay coefficient is 0.1;

[0053] The classification network model is a deep learning framework based on PyTorch, and the TensorRT inference engine is used to accelerate the inference of the classification network model;

[0054] S4, deploying the trained classification network model on the edge computing device 4, building an online monitoring system in combination with the image acquisition platform 3, and performing real-time monitoring of the vibrating screen;

[0055] The core module of edge computing device 4 is NVIDIA's Jetson Xavier NX embedded artificial intelligence computing processor, forming the LCFC edge intelligent computer EA-B310.

[0056] S5. Process and analyze the collected machine-made sand image in real time. When the system determines that the screen is damaged and the particle size in the machine-made sand image reaches a specified threshold, the system issues a screen damage alarm;

[0057] The images identified as mixed materials and normal single-grade material images are counted, and the proportion is calculated, and a proportion threshold is set. Specifically, the first count is performed when the model identifies the image as a mixed material, and the mixed material images are continuously counted and the normal single-grade material machine-made sand images are collected within a specified time segment. When the proportion of mixed material images to normal machine-made sand images reaches the set threshold, a screen damage alarm is issued. If the set threshold is not reached, a new time segment and image count is performed.

[0058] The specific manifestations of the screen damage alarm issued by the system include: displaying the current machine-made sand image, prompt information displayed synchronously with the machine-made sand image, and the sound and light alarm 5 running synchronously with the sound and light alarm information;

[0059] The monitoring system issues a screen damage alarm, the vibrating screen receives the fault signal and shuts down, and the human-computer interaction interface displays which screen is damaged, prompting you to replace the screen.

[0060] According to some embodiments of the present invention, optionally, Figure 4 As shown, the present invention also provides a vibrating screen damage monitoring device based on deep learning, comprising:

[0061] The screening device 1 has several vibrating screen meshes of different densities arranged therein;

[0062] A material discharge chute 2, which is connected to the discharge port of the screening device 1, and the material discharge chute 2 is inclined;

[0063] An image acquisition platform 3 is disposed above the material discharge chute 2, and the image acquisition platform 3 is perpendicular to the material discharge chute 2;

[0064] An edge computing device 4, which is electrically connected to the image acquisition platform 3;

[0065] The sound and light alarm 5 is arranged on the screening device 1 and is electrically connected to the edge computing device 4 .

[0066] Specifically, the image acquisition platform 3 is composed of a detection box, an industrial camera, a lens and an LED screen flash source; the purpose of being installed in the inclined section of the unloading chute 2 is that the falling speed of the machine-made sand in this part is relatively slow, which can meet the exposure requirements of the industrial camera.

[0067] According to some embodiments of the present invention, optionally, a distributor 6 is provided in the discharge chute 2 at a position corresponding to the image acquisition platform 3 , and the distributor 6 raises the height of the machine-made sand falling into the discharge chute 2 .

[0068] The built-in distributor 6 is a mechanical structure device for sampling and diverting the machine-made sand in the chute, the purpose of which is to raise the height of the machine-made sand to meet the depth of field limitation of the industrial camera.

[0069] According to some embodiments of the present invention, optionally, the edge computing device 4 includes a memory and a processor, the memory stores a computer program for monitoring the damage of a vibrating screen, and the computer program is loaded and executed on the processor.

[0070] It should be understood that the embodiments disclosed in the present invention are not limited to the specific processing steps or materials disclosed herein, but should be extended to equivalent substitutions of such features understood by ordinary technicians in the relevant field. It should also be understood that the terms used herein are only used for the purpose of describing specific embodiments and are not meant to be limiting.

[0071] The "embodiment" mentioned in the specification means that the specific features or characteristics described in conjunction with the embodiment are included in at least one embodiment of the present invention. Therefore, the phrases or "embodiment" appearing in various places throughout the specification do not necessarily refer to the same embodiment.

[0072] In addition, the described features or characteristics may be combined in one or more embodiments in any other suitable manner. In the above description, some specific details, such as thickness, quantity, etc., are provided to provide a comprehensive understanding of the embodiments of the present invention. However, those skilled in the relevant art will understand that the present invention can be implemented without one or more of the above specific details or can also be implemented using other methods, components, materials, etc.

Claims

1. A vibration screen damage monitoring method based on deep learning, characterized in that: include: S1. Using a vibrating screen to screen the machine-made sand, the machine-made sand after screening falls into the material discharge chute below, and using an image acquisition platform to collect images of the machine-made sand in the material discharge chute. The machine-made sand images specifically include: images of single-grade materials before the screen is damaged, and images of mixed materials of machine-made sand with a larger particle size after the screen is damaged; S2, screening the obtained machine-made sand images and making a binary classification data set; S3, input the binary classification data set into the lightweight convolutional neural network for training to obtain a classification network model; S4. Deploy the trained classification network model on the edge computing device, build an online monitoring system in combination with the image acquisition platform, and monitor the vibrating screen in real time; S5. Process and analyze the collected machine-made sand images in real time. When the system determines that the screen is damaged and the particle size in the machine-made sand image reaches a specified threshold, the system issues a screen damage alarm; count the images determined as mixed materials and normal single-grade material images, calculate the proportion, and set the proportion threshold. Specifically, when the model determines that the image is a mixed material, the first count is performed, and the mixed material images and normal single-grade material machine-made sand images are continuously counted within a specified time segment. When the proportion of the mixed material image to the normal machine-made sand image reaches the set threshold, the screen damage alarm is issued. If the set threshold is not reached, a new time segment and image count is performed; The machine-made sand screened by the complete vibrating screen is machine-made sand of a single particle size grade, the material discharge chute is inclined, and the image acquisition platform is arranged perpendicularly to the material discharge chute; In the S3 step, the lightweight convolutional neural network adopts the MobileNetV3 network structure. The large version includes 15 bottleneck layers, one standard convolution layer, and three point-by-point convolution layers. The bottleneck layer contains multiple depth-separable convolution layers. The classification network model is a deep learning framework based on PyTorch, and the TensorRT inference engine is used to accelerate the inference of the classification network model.

2. The vibrating screen damage monitoring method based on deep learning according to claim 1 is characterized in that: In the process of converting the machine-made sand image into a binary classification data set in step S2, a data enhancement step is also required. The specific methods of data enhancement include: horizontal flipping, vertical flipping, brightness change, adding noise and blurring.

3. The vibrating screen damage monitoring method based on deep learning according to claim 1 is characterized in that: In step S4, the core module of the edge computing device is the NVIDIA Jetson Xavier NX embedded artificial intelligence computing processor.

4. The vibrating screen damage monitoring method based on deep learning according to claim 1 is characterized in that: In step S5, the specific manifestations of the system issuing a screen damage alarm include: displaying the current machine-made sand image, prompt information displayed synchronously with the machine-made sand image, and sound and light alarm information running synchronously with the sound and light alarm.

5. A vibrating screen damage monitoring device based on deep learning, using the vibrating screen damage monitoring method based on deep learning as described in any one of claims 1 to 4, characterized in that: include: A screening device, in which a plurality of vibrating screens of different densities are arranged; A material discharge chute connected to the discharge port of the screening device, and the material discharge chute is inclined; An image acquisition platform is disposed above the material discharge chute, and the image acquisition platform is perpendicular to the material discharge chute; An edge computing device electrically connected to the image acquisition platform; An audible and visual alarm is arranged on the screening device and is electrically connected to the edge computing device.

6. The vibrating screen damage monitoring device based on deep learning according to claim 5 is characterized in that: In the material discharge chute, a material distributor is arranged at a position corresponding to the image acquisition platform, and the material distributor raises the height of the machine-made sand falling into the material discharge chute.

7. The vibrating screen damage monitoring device based on deep learning according to claim 6 is characterized in that: The edge computing device includes a memory and a processor. The memory stores a computer program for monitoring the damage of a vibrating screen, and the computer program is loaded and executed on the processor.

Images